Page 1 of 1

Estimating and forecasting for one day with no equal sample

Posted: Tue Mar 10, 2015 5:33 am
by Afterall
I have a series of €/ruble exchange rate and oil price for 2014. And I want to test and forecast an equation: rube=b1+b2*oil+b3*ruble(-1)
I have the sample time from 1/1/2014 to 3/31/2014 and I did a forecast for 4/1/2014.
But Now I want to estimation the equation with sample 1/1/2014 to 4/1/2014 and do a forecast for 4/2/2014 and save to rublef.
And then estimate with sample 1/1/2014 to4/2/2014 and forecast 4/5/2014 and save this value to rublef.

So, the main problem: How to estimate and forecast with different samples and save forecasted value to new forecasted serie. I think this can be solved with some FOR cases, but I do`t know how.
I hope anyone know this.

Re: Estimating and forecasting for one day with no equal sam

Posted: Tue Mar 10, 2015 5:37 am
by trubador
You can (re)search for rolling forecast.

Re: Estimating and forecasting for one day with no equal sam

Posted: Wed Mar 11, 2015 2:48 am
by Afterall
Thank you. I didn't know the right word for those kind of forecasting. I've found the code and idea and did the program, and works correct.
If will anyone search for this problem, here is all:
ftp://ftp.physics.uoc.gr/pub/tmp/Panago ... s/roll.htm